DISOselect

DISOselect estimates the expected performance of 12 intrinsic disorder predictors for individual protein sequences using sequence-derived properties to guide selection of the most appropriate predictor for intrinsic disorder analysis.

Key Features:

• Predictor Performance Estimation: Evaluates the expected quality of predictions from a selection of 12 intrinsic disorder predictors for individual proteins.
• Sequence-Driven Analysis: Relies solely on sequence-derived properties and operates independently of existing disorder prediction outputs.
• Input Format: Operates on protein sequences provided in FASTA format.
• Empirical Validation: Selection by DISOselect produced statistically significant improvements in predictive performance on a test set of 1,000 proteins.

Scientific Applications:

• IDP characterization: Guides selection of disorder predictors to improve analysis of intrinsically disordered proteins (IDPs).
• Functional inference: Improves reliability of disorder-based insights into protein function.
• Interaction and target analysis: Supports more accurate mapping of disorder-related interaction networks and identification of potential therapeutic targets.

Methodology:

DISOselect uses sequence-derived properties to estimate per-protein predictor performance for a panel of 12 intrinsic disorder predictors and makes selection recommendations without requiring prior disorder predictions.